This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Optimized estimator for real-time dynamic displacement
measurement using accelerometers
Jonathan Abir a , ∗, Stefano Longo
b , Paul Morantz
a , Paul Shore
c
a The Precision Engineering Institute, School of Aerospace, Transport and Manufacturing, Cranfield University, Cranfield, MK43 0AL, UK b The Centre for Automotive Engineering and Technology, Cranfield University, Cranfield, MK43 0AL, UK c The National Physical Laboratory, Teddington, TW11 0LW, UK
a r t i c l e i n f o
Article history:
Received 19 May 2016
Revised 11 July 2016
Accepted 22 July 2016
Keywords:
Accelerometer
Displacement estimation
Flexible frame
Heave filter
Virtual metrology frame
a b s t r a c t
This paper presents a method for optimizing the performance of a real-time, long term, and accurate ac-
celerometer based displacement measurement technique, with no physical reference point. The technique
was applied in a system for measuring machine frame displacement.
The optimizer has three objectives with the aim to minimize phase delay, gain error and sensor noise.
A multi-objective genetic algorithm was used to find Pareto optimal estimator parameters.
The estimator is a combination of a high pass filter and a double integrator. In order to reduce the
gain and phase errors two approaches have been used: zero placement and pole-zero placement. These
approaches were analysed based on noise measurement at 0g-motion and compared. Only the pole-zero
placement approach met the requirements for phase delay, gain error, and sensor noise.
Two validation experiments were carried out with a Pareto optimal estimator. First, long term mea-
surements at 0g-motion with the experimental setup were carried out, which showed displacement error
of 27.6 ± 2.3 nm. Second, comparisons between the estimated and laser interferometer displacement mea-
surements of the vibrating frame were conducted. The results showed a discrepancy lower than 2 dB at
t contains 16 I/O channels and 16 bit Analog to Digital Converter
ADC). The conversion time for each ADC is 5 μs. The target ma-
hine is optimized for MathWorks ® SIMULINK
® and xPC Target TM
.
The frame displacement x f was estimated by measuring frame
ibration a f using low noise accelerometers; the signal was ac-
uired by the ADC and passed through the estimator. It is com-
osed of a High Pass Filter (HPF) to reduce low frequency noise
nd a numerical double integrator ( Fig. 7 ).
Laser interferometer (Renishaw ML10 Gold Standard) was used
o validate the estimated displacement. The laser light is split into
wo paths by a beam splitter, one that is reflected by a “dynamic”
etroreflector and another reflected by a “stationary” retroreflec-
or ( Fig. 8 a). The dynamic retroreflector was mounted to the ma-
hine frame, and an accelerometer mounted to the retroreflector
Fig. 8 b). The stationery retroreflector was fixed using an optics
older. Note that this validation setup can only be realized on the
implified motion system, and not on the full machine. The dis-
lacement of the dynamic retroreflector is measured by counting
he number of interference events. The interferometer can mea-
ure dynamic displacement at a sampling rate of 50 0 0 Hz with a
esolution of 1 nm.
.2. The optimization problem
Real-time implementation of displacement measurements
ased on integration in a control system requires small phase
elay, low gain error, and high sensor noise removal. The multi-
bjective optimization problem can be formally defined as: find
he vector � x = [ � x 1 , � x 2 , . . . , � x n ] T , which satisfies n constrains (2)
i
(→
x
)≥ 0 , i = 1 , . . . , n, (2)
nd optimize the vector function
�
( � x ) = [ � J σ ( � x ) , � J Mag ( � x ) , � J Phase ( � x )] T , (3)
here � x is the estimator parameters vector that simultaneously
inimizes the three error functions: noise error function J σ , mag-
itude error function J Mag , and phase error function J Phase .
. Displacement estimation noise analysis
This section presents the noise sources in the acceleration mea-
urement, and the effect of acceleration noise on the displacement
stimation.
4 J. Abir et al. / Mechatronics 39 (2016) 1–11
(b)(a)
Carriage
Frame
EncoderMotion
axis
Fig. 5. Experimental modal analysis results. Unstressed frame (a), and a flexible frame mode shape at 305 Hz (b). The length of the arrows in (b) represents the mode shape
magnitude.
AccelerometersFrameEncoder scale
Air bearingEncoder readhead
Fig. 6. Setup of four IEPE accelerometers mounted to the machine frame.
DSP
Integrator
Analog to Digital Converter High Pass Filter
Signal conditioner Accelerometer
xf
af
Fig. 7. Estimating displacement x f by acceleration signal a f block diagram.
C
∼
i
n
e
a
A
w
u
n
a
l
t
t
1
f
q
r
t
p
o
H
r
f
4
t
t
f
d
a
w
t
i
p
There are three uncorrelated sources that contribute to the dis-
placement measurement noise ( Fig. 9 ): accelerometer, signal con-
ditioner and ADC. The Power Spectral Density (PSD) of each source
is usually specified by the manufacturer. The accelerometer has the
lowest noise contribution; however, to improve the Signal to Noise
Ratio (SNR) a x100 gain is used. Thus, the accelerometer noise is
the most significant noise source.
It is difficult to interpret the noise contribution of a specific
bandwidth from a PSD graph, which can be circumvented by cal-
culating the Cumulative Power Spectrum (CPS) [27] :
P S( f ) =
∫ f e
f s
P SD (υ) / ( 2 π · υ) 4 d υ. (4)
Note that the sensor PSD is in acceleration units and the CPS
is in displacement, thus the PSD is multiplied by (2 π · f ) −4 . The
CPS graph ( Fig. 10 ) shows that the expected displacement noise is
10 μm mainly due to low frequency noise < 5 Hz; however, this
s an optimistic analysis since it was not considered low frequency
oise ( < 1 Hz).
In [17,28] an empirical formula was suggested for resolution
stimation of the displacement measurement A based on the
ccelerometer spectral density ρ and measurement time T m
:
¯ (ρ, T m
) = η
(∫ (1+ ν) / T m
(1 −ν) / T m
ρ( f ) df
)0 . 5
T 2 m
, (5)
here ν is experimentally determined constant, and η depending
pon signal acquisition and conditioning [29] . The displacement
oise resembles a sine wave, whose frequency is close to T m
−1 ,
nd its amplitude is A .
Eq. (5) can be transformed to estimate the maximum allowance
ow-frequency noise ρLF for a given resolution A and measurement
ime T m
:
ρLF =
(2 π2 · A
)2
ν · T 3 m
. (6)
For example, to achieve the resolution of A = 30 nm for dura-
ion of T m
= 20 s the low frequency noise should be ρLF ≤ 1 . 75 ·0 −4 ( μm / s 2 ) 2 / Hz . This low noise requirement is technologically
easible [26,30–32] , but there is no information at very low fre-
uencies required for long duration measurements [28] .
Based on the noise analysis, the displacement estimator must
educe the low frequency noise significantly as there are at least
en orders of magnitude difference between the required and ex-
ected noise level. By plotting the CPS (4) as a function of f s ne can assess the minimum required cut-off frequency for an
PF ( Fig. 11 ). By finding the intersection of the CPS line with the
equired noise level, 30 nm, the minimum cut-off frequency was
ound to be 17 Hz.
. Estimator design
This section presents an estimator design based on a combina-
ion of a high pass filter and a double integrator, and phase correc-
ion techniques.
An accelerometer can be regarded as a single-degree-of-
reedom mechanical system, with a simple mass, spring, and
amper [33] . Its output signal can be represented as [34] :
(t) = k a x (t) + w (t) + w 0 , (7)
here k a is the accelerometer gain, x is the acceleration acting on
he accelerometer, w(t) is the noise and disturbance effect, and w 0
s the 0g-offset.
Although integration is the most direct method to obtain dis-
lacement from acceleration, due to 0g-offset and low frequency
J. Abir et al. / Mechatronics 39 (2016) 1–11 5
(a)
Environmentalcompensation unit
Optics holder
Dynamic retroreflector
Dynamic accelerometer
Stationaryretroreflector
Beam splitter
Carriage motion
Laser sourcedetectors
Stationaryretroreflector
Dynamicretroreflector
Beamsplitter
Dynamicaccelerometer
Displacement
(b)
Fig. 8. Setup of the displacement validation experiment.
Fig. 9. Power Spectral Density of accelerometer multiplied by 100x gain PSD a , sig-
nal conditioner PSD sg , ADC PSD adc and the total Power Spectral Density PSD t .
Fig. 10. Cumulative Power Spectrum of accelerometer multiplied by 100x gain, sig-
nal conditioner, ADC, and total Cumulative Power Spectrum, CPS a , CPS sg , CPS adc , and
CPS t respectively. f s = 1 Hz as PSD is not specified for lower frequencies.
Fig. 11. CPS as a function of f s .
n
r
M
T
c
b
p
t
[
f
a
t
H
o
b
h
i
m
d
f
f
oise it is not appropriate to integrate the acceleration signal di-
ectly. The integration process leads to an output that has a Root
ean Square (RMS) value that increases with integration time [20] .
his can be a problem even in the absence of any motion of the ac-
elerometer, due to the 0g-offset [18,35] . Displacement estimation
ased on digital integration was shown to have lower noise com-
ared to analog integration. Furthermore, at high sampling rates
he digital integration showed higher accuracy [16] .
Numerical integrators can be used in the time domain
20,36,37] and in the frequency domain [15,38] ; however, using
requency domain techniques for real time application is difficult,
s it suffers from severe discretization errors, if the discrete Fourier
ransform is performed on a relatively short time interval [15,39] .
ence, the digital estimator will be in the time domain.
A HPF is used to remove constant or low frequency offsets (0g-
ffset) and to reduce the low frequency noise. Without it, the dou-
le integrated signal will diverge due to the double integration be-
avior. Tuning the HPF, an optimized cut-off frequency, ω c , takes
nto consideration good tracking of the actual displacement, re-
oval of sensor noise and offsets, and low phase errors [40] . Re-
uced gain and phase lead error are associated with a high cut-off
requency, whereas high noise gain is associated with a low cut-off
requency.
6 J. Abir et al. / Mechatronics 39 (2016) 1–11
Fig. 12. Comparison of heave filters and ideal double integrator Bode plots. H dbl is an ideal double integrator. H heave , H ZPF , and H PZP are the heave filter, heave filter with
zero-placement filter, and heave filter with pole-zero-placement filter respectively, where ω c = 17 Hz and ς = 1/ �2.
G
4
p
G
g
s
a
o
t
e
c
i
c
w
t
F
s
o
�x
�x
5
t
t
d
J
w
i
c
“
0
3
g
f
t
An estimator based on a combination of an HPF and a double
integrator (heave filter) is given by [40] :
H hea v e =
ˆ D
a (s ) =
s 2
( s 2 + 2 ς ω c + ω c 2 )
2 , (8)
where s is the Laplace variable, ς the damping coefficient, and
ω c the cut-off frequency of the filter. The displacement estimate
is denoted by ˆ D . Usually ς= 1/ �2, to obtain a Butterworth con-
tour. It was used for estimating the heave position of a ship due
to sea waves [40,41] , and displacement of optical elements due to
structural vibrations [22] using accelerometer. The drawback of the
heave filter is phase errors, which can be reduced by using an All
Pass Filter (APF), Zero Placement Filter (ZPF) or Pole-Zero Place-
ment (PZP).
The APF can eliminate the phase error for one specific fre-
quency thus it is not applicable for our case. In ZPF (9) , one of the
zeros z a is placed to reduce significantly the phase error; however,
it attenuates noise less at low frequencies [41] . In PZP (10) , a pole
p and zero z are added to the heave filter with an additional gain
parameter K .
H ZPF =
s (s + z a )
( s 2 + 2 ς ω c + ω c 2 )
2 (9)
H PZP =
s 2
( s 2 + 2 ς ω c + ω c 2 )
2 · K
s − z
s − p (10)
In Fig. 12 a comparison of heave filters with phase correction
filter and ideal double integrator (11) is shown, where the mini-
mum cut-off frequency ω c = 17 Hz was based on the displacement
noise analysis ( Fig. 11 ). As can be seen, the filters magnitude above
the cut-off frequency is as a double integrator; however, the phase
differs significantly from the ideal −180 º phase even for frequen-
cies > 3 �ω c .
H dbl =
ˆ D
a (s ) =
1
s 2 (11)
5. Optimizer design
This section describes the optimizer design, its constraints and
goals, and the three error functions. The design is independent of
the estimator design.
During the last decade, it was shown that evolutionary algo-
rithms are useful in solving multi-objective optimization problems.
There are various techniques which can be used, however the Ge-
netic Algorithm (GA) approach was chosen. It has been shown that
A is suitable for solving complex mechatronics problems [42–
4] especially for signal processing [45] , and for multi-objectives
roblems [46,47] .
As there are three objectives ( J σ , J Phase , J Mag ) a Multi-Objective
enetic Algorithm (MOGA) optimizer was chosen using Matlab
lobal optimization toolbox [48] . The algorithm scans the whole
earch domain and exploits promising areas by selection, crossover
nd mutation applied to individuals in population. In multi-
bjective optimization, the aim is to find good compromises rather
han a single solution. An optimal solution
� x ∗ is found if there
xists no feasible � x which would decrease one objective without
ausing a simultaneous increase in at least one other objective. The
mage of the optimal solutions is called the “Pareto Front”. The de-
ision maker chooses a solution from the Pareto optimal solutions
hich compromise and satisfies the objectives as possible. A de-
ailed explanation of the GA process can be found in [45–47,49] .
or three objective functions it gives a surface in three-dimensional
pace. Thus, the “optimal” solution is chosen by the designer.
The vector � x is defined by the estimator parameters in the cases
f ZPF (12) and PZP (13) . To obtain a Butterworth contour ς= 1/ �2.
= [ � ω c , � z a ] T (12)
= [ � ω c , � K , � z , � p ] T (13)
.1. Error functions
As described in Section 2 , there are three displacement estima-
ion error functions: noise error function J σ , magnitude error func-
ion J Mag , and phase error function J Phase .
The noise error function is defined as the RMS value σ ˆ D of the
isplacement signal generated by the estimator:
σ · T σ = σ ˆ D , (14)
here T σ is noise normalization factor.
In section 3 it was shown that low frequency acceleration noise
s the main contributor to J σ , however usually there is no specifi-
ation for noise at these frequencies. Hence, for this calculation a
0g-motion” acceleration noise measurements were used. A typical
g-motion acceleration signal was measured at a sampling rate of
2 kHz for t = 20 s. The measured CPS and the expected CPS are in
ood agreement as can be seen in Fig. 11.
The magnitude and phase error functions are defined as the dif-
erence between the magnitude and phase response of the estima-
or and an ideal double integrator. The magnitude error function
J. Abir et al. / Mechatronics 39 (2016) 1–11 7
i
J
w
d
w
f
J
w
i
t
e
t
v
5
t
t
b
e
J
J
J
b
q
s
t
q
1
m
z
0
0
h
c
−
6
f
Fig. 13. Pareto front of H ZPF when ς = 1/ �2.
6
c
p
o
e
e
q
h
w
u
l
i
t
H
T
h
6
c
t
s
a
U
H
a
l
c
K
t
a
s:
Mag · T Mag =
n ∑
i =1
| M dbl ( f i ) − M HF ( f i ) | · w i , (15)
here M dbl (f i ) and M HF (f i ) are the magnitude response of an ideal
ouble integrator and the estimator at frequency f i , respectively,
i is a weighting vector, and T Mag is the magnitude normalization
actor. The phase error function is:
Phase · T Phase =
n ∑
i =1
| P dbl ( f i ) − P HF ( f i ) | · w i , (16)
here P dbl (f i ) and P HF (f i ) are the phase response of the ideal double
ntegrator and the estimator at frequency f i , respectively. T Phase is
he phase normalization factor.
For simplicity the frequencies f i , that were used to calculate the
rror functions, are the frame resonances which are obtained from
he plant FRF ( Fig. 4 ); however, one can use a different frequency
ector (and it corresponding weighing vector w i ).
.2. Optimizer constraints and goals
The goals for the noise and phase errors are set due to sys-
em requirements. The noise level is required to be comparable to
he linear encoder noise (17) , and the phase error is required to
e smaller than the servo control update rate (18) . The magnitude
rror was set empirically (19) .
σ < 30 nm (17)
Phase < 70 μs (18)
Mag < 3 dB (19)
The maximum constraint to the cut-off frequency ω c is defined
y the first antiresonance frequency ( Fig. 4 ), as above that fre-
uency frame vibrational displacement occurs. The minimum con-
traint to the cut-off frequency was calculated in Section 3 using
he CPS plot ( Fig. 10 ). Hence, the constraints for the cut-off fre-
uency are
7 · 2 π rad / s ≤ ω c ≤ 75 · 2 π rad / s . (20)
In the case of ZPF, the maximum constraint for the zero place-
ent z a is given [40] . The minimum constraint is that it has no
ero placement hence,
< z a ≤ 2
√
2 · ω c = 1332 . 9 rad / s . (21)
In the case of PZP, the gain factor K has the constraints of
< K ≤ 1 . (22)
There is no analytical equation for the pole and zero values;
owever, one should consider a stable filter where
z
p > 1 . (23)
Thus, an empirical approach was used to set the pole and zero
onstraints:
30 0 0 ≤ z, p < 0 . (24)
. Results of the optimization process
This section presents the optimization results and its Pareto
ront graphs for the two estimator designs ZPF and PZP.
.1. Zero placement filter
The Pareto front graph of the ZPF estimator with Butterworth
ontour is shown in Fig. 13 . The main conflicting objectives are the
hase error and noise level. No optimal solution which meets the
ptimization goals (17–19) was found. Allowing an underdamped
stimator, where 0 < ς ≤ 1/ �2, improves the phase response at the
xpense of having a resonant peak. Hence, a higher cut-off fre-
uency is possible which reduces the sensor noise and phase error;
owever, no optimal solution which meets all three requirements
as found.
Thus, a ZPF estimator design with either Butterworth contour or
nderdamp properties is not an appropriate solution for the prob-
em.
A new approach was proposed by splitting the denominator
nto two different filters, which adds two degrees of freedom to
he ZPF design (25) .
¯ ZPF =
s · ( s + z a ) (s 2 + 2 ς 1 ω c1 + ω c1
2 )
·(s 2 + 2 ς 2 ω c2 + ω c2
2 ) (25)
Hence, its new parameters vector is � x = [ � ω c1 , � ω c2 , � ς 1 , � ς 2 , � z a ] T .
his new approach shows improvement of the Pareto front results
owever, no results were found which meets the requirements.
.2. Pole-zero placement filter
The Pareto front graph of the PZP estimator with Butterworth
ontour is shown in Fig. 14 . Again in this case there is no solu-
ion which meets the requirements, although it shows better re-
ults compared to Fig. 13.
Allowing an underdamped estimator, where 0 < ς ≤ 1/ �2 gives
n optimal solutions which meets the set of requirements ( Fig. 15 ).
nd parameter vector � x = [ � ω c1 , � ω c2 , � ς 1 , � ς 2 , � K , � z , � p ] T , shows simi-
ar results. Thus, based on Fig. 15 an optimized estimator was
hosen with the following parameters: ς = 0.37 , ω c = 214.82 rad/s,
= 0.69, z = − 429.29, p = − 11.90.
Comparing the Butterworth contour and underdamped H PZP
ransfer functions ( Fig. 16 ) emphasizes that a lower damping ratio
llows higher cut-off frequency at the expense of a resonant peak.
8 J. Abir et al. / Mechatronics 39 (2016) 1–11
Fig. 14. Pareto front of H PZP when ς = 1/ �2.
Fig. 15. Pareto front of H PZP when 0 < ς ≤ 1/ �2.
a
n
7
t
v
(
o
i
7
m
s
n
a
v
m
s
t
a
i
f
a
7
s
p
f
a
f
c
h
f
o
t
t
F
f
Above the cut-off frequency the ZPF design has a positive gain er-
ror, while in PZP designs it changes its sign. Moreover, a significant
difference in the low frequency noise reduction between the ZPF
Fig. 16. Comparison of optimized heave filters and the ideal double integrator. H dbl is
placement filter and with pole-zero-placement filter respectively.
nd PZP designs can be observed. The PZP designs have a better
oise reduction.
. Estimator validation
This section shows the experimental results validating an op-
imal PZP estimator design. There are three main experiments for
alidating the estimator performance: (i) robustness of the design;
ii) long term measurements at 0g-motion; and (iii) a comparison
f displacement signals due to structural vibrations between laser
nterferometer sensor and the displacement based acceleration.
.1. 0g-motion noise and robustness validation
Using the optimized estimator, 0g-motion measurement was
ade with four tri-axial accelerometers (12 accelerometers) as
hown in Fig. 6 . The setup is detailed in Section 2.1 . The sig-
als were acquired at a sampling rate of 54 kHz for t = 600 s. The
chieved displacement RMS is 27.6 ± 2.3 nm. Furthermore, the low
ariance between all of the accelerometers assures that the esti-
ator design is robust, and not accelerometer dependent. Fig. 17 a
hows the estimated displacement of 0g-motion measurement, i.e.
he RMS noise, of one typical accelerometer. The results are in
greement with the requirements (17) . Fig. 17 b shows the changes
n displacement noise RMS over measurement time. As required
rom the estimator, 0g-offset and low frequency noise are attenu-
ted which allows long term double integration without diverging.
.2. Displacement estimation validation
The validation was made by comparing the displacement mea-
ured by the laser interferometer and the acceleration based dis-
lacement measurement of the machine frame vibrations. The
rame was excited using an oscillating position command gener-
ted by the linear motion controller, X set = A i ·sin( ω i t ), at various
requencies ω i and amplitudes A i ( Table 1 ). Note that A i is the
ommanded carriage movement amplitude; hence the frame ex-
ibits different displacement due to the servo reaction forces. The
rame displacement amplitudes measured by the laser interfer-
meter ( A L,i ) and acceleration based displacement ( A est,i ) were ex-
racted using a Fast Fourier Transform (FFT). The discrepancy be-
ween the measurements meets the specified requirements (19) .
ig. 18 shows an example of the discrepancy in the measured
rame displacement at 100 Hz.
an ideal double integrator. H ZPF and H PZP are optimized heave filters with zero-
J. Abir et al. / Mechatronics 39 (2016) 1–11 9
Fig. 17. Displacement estimation noise measurement. (a) Noise in long term measurement for t = 600 s. (b) Noise Root Mean Square (RMS) of the displacement signal.
Table 1
Results of displacement estiamtion validation.
i ω i [Hz] A i [nm] A L,i [nm] A est,i [nm] Discrepancy [dB]
1 80 40 ,0 0 0 1533 .0 1932 .0 2 .0
2 100 15 ,0 0 0 741 .0 751 .0 0 .1
3 100 20 ,0 0 0 981 .6 998 .9 0 .1
4 120 14 ,0 0 0 292 .4 261 .2 0 .9
5 150 20 ,0 0 0 263 .2 232 .2 1 .0
6 200 20 ,0 0 0 93 .5 78 .6 1 .5
Fig. 18. Displacement estimation validation at X set = 15,0 0 0 ·sin(10 0 ·2 π ·t ). A L,i and
A est,i are the frame displacement amplitude measured by the laser interferometer
and acceleration based displacement respectively.
8
s
s
d
a
o
r
t
i
b
c
m
c
b
e
n
f
d
m
m
w
A
E
P
g
a
. Conclusions
This research shows that accelerometers can be used to mea-
ure real-time displacement in the nanometer range without con-
traints to the integration time.
Common displacement sensors require a reference point, which
oes not always exist. Thus, the novelty of this technique is the
bility to measure the dynamic displacement of a structure with-
ut having a physical reference point, but instead using a “virtual”
eference point. Doing so, it was assumed that the initial condi-
ions of the frame is unstressed state and in rest. The feasibil-
ty of this technique depends on the lowest frequency required to
e measure since that low frequency noise is the most significant
ause of displacement error. Although the displacement noise and
easurement bandwidth met the requirements, by using an ac-
elerometer with higher performance the displacement noise can
e reduced significantly and the measurement bandwidth can be
xtended towards 0 Hz. Furthermore, using acceleration based dy-
amic displacement measurement technique offers an unlimited
ull-scale-range sensor in the nanometer range.
The optimized estimator showed less than 10% variation in the
isplacement noise with different accelerometers (from the same
odel) which demonstrate its robustness.
The developed technique is essential to realize the virtual
etrology frame concept. Thus, it was implemented in a machine
ith a flexible frame improved it dynamic performance.
cknowledgement
This work was supported by the UK EPSRC under grant
P/I033491/1 and the Centre for Innovative Manufacturing in Ultra-
recision. The author is grateful to the McKeown Precision En-
ineering and Nanotechnology foundation at Cranfield University,
nd B’nai B’rith Leo Baeck (London) for their financial support.
10 J. Abir et al. / Mechatronics 39 (2016) 1–11
[
References
[1] Huo D , Cheng K , Wardle F . Int J Adv Manuf Technol 2010;47:867–77 .
[2] Brecher C , Utsch P , Klar R , Wenzel C . Int J Mach Tools Manuf 2010;50:328–34 .
[3] Shore P , Morantz P , Read R . 10th Int. conf. exhib. laser metrol. mach. tool, c.robot. performance., euspen; 2013 .
[4] H Soemers, Design principles: for precision mechanisms, T-Pointprint, 2011. [5] Butler H . IEEE Control Syst Mag 2011;31:28–47 .
[6] Fan KC , Fei YT , Yu XF , Chen YJ , Wang WL , Chen F , et al. Meas Sci Technol2006;17:524–32 .
[7] Takahashi M , Yoshioka H , Shinno H . J Adv Mech Des Syst Manuf
2008;2:356–65 . [8] Leadbeater PB , Clarke M , Wills-Moren WJ , Wilson TJ . Precis Eng 1989;11:191–6 .
[9] Coelingh E , De Vries TJA , Koster R . IEEE/ASME Trans Mechatron2002;7:269–79 .
[10] Rankers AM , van Eijk J . In: Proc. second int. conf. motion vib. control. yoko-hama; 1994. p. 711–16 .
[11] Abir J , Morantz P , Shore P . In: Proc. 15th int. conf. eur. soc. precis. eng. nan-otechnol.; 2015. p. 219–20 .
[12] Abir J , Morantz P , Longo S , Shore P . In: ASPE 2016 spring top. meet. precis.
mechatron. syst. des. control, american society for precision engineering, ASPE;2016. p. 58–61 .
[13] Fleming AJ . Sens Actuators A 2013;190:106–26 . [14] Gao W , Kim SW , Bosse H , Haitjema H , Chen YL , Lu XD , et al. CIRP Ann Manuf
Technol 2015;64:773–96 . [15] Ribeiro JGT , De Castro JTP . IMAC-XXI A conf. expo. struct. dyn.; 2003 .
[33] Levinzon F . Piezoelectric accelerometers with integral electronics. Springer;2015 .
[34] Zhu W-H , Lamarche T . Ind Electron IEEE Trans 2007;54:2706–15 . [35] Thenozhi S , Yu W , Garrido R . Trans Inst Meas Control 2013;35:824–33 .
[36] Gavin HP , Morales R , Reilly K . Rev Sci Instrum 1998;69:2171 .
[37] Razavi SH , Abolmaali A , Ghassemieh M . Comput Methods Appl Math2007;7:227–38 .
[38] Worden K . Mech Syst Signal Process 1990;4:295–319 . [39] Lee HS , Hong YH , Park HW . Int J Numer Methods Eng 2011;82:1885–91 .
[40] Godhaven J-M . In: IEEE ocean. eng. soc. ocean. conf. proc.. IEEE; 1998. p. 174–8 .[41] Richter M , Schneider K , Walser D , Sawodny O . In: Int. fed. autom. control world
congr.; 2014. p. 10119–25 .
[42] Lin C-L , Jan H-YJ , Shieh N-C . IEEE/ASME Trans Mechatron 2003;8:56–65 . [43] Ito K , Iwasaki M , Matsui N . IEEE/ASME Trans Mechatron 2001;6:143–8 .
44] Van Brussel H , Sas P , Németh I , De Fonseca P , Van Den Braembussche P .IEEE/ASME Trans Mechatron 2001;6:90–105 .
[45] Tang KS , Man KF , Kwong S , He Q . IEEE Signal Process Mag 1996;13:22–37 . [46] Deb K . Multi-Objective optimization using evolutionary algorithms. John Wiley
& Sons; 2001 .
[47] Coello CC , Lamont GB , van Veldhuizen DA . Evolutionary algorithms for solvingmulti-objective problems. Springer Science & Business Media; 2007 .
[48] Mathworks. Global optimization toolbox, user’s guide, version 3. Mathworks;2015 .
[49] Man KF , Tang KS , Kwong S . IEEE Trans Ind Electron 1996;43:519–34 .
J m Ben-Gurion University, Beer-Sheva, Israel in 2008, and a M.Sc. degree in mechanical e d a Ph.D. degree at the Center for Innovative Manufacturing in Ultra-Precision, Cranfield
U elopment of opto-mechanical systems for the defense sector, where he was awarded for h
P e at Cranfield University. He is an expert in precision metrology and has devised the
m oss a wide range of platforms. He has extensive academic research interests and industrial e
S Systems at Cranfield University, a Senior Member of the IEEE and a Chartered Engineer.
H vements in the area of Control and Automation” and an Honorary Research Associate at I ity of Sheffield, UK, in 2007 and completed his Ph.D. in control systems at the University
o arch associate at Imperial College London. He joined Cranfield University in the summer o
P to studying at Liverpool Polytechnic and then undertaking MSc and Ph.D. in Precision
E d for SKF Group in the Netherland and Sweden where he managed the development of
p rofessor of Ultra Precision Technologies and headed their Precision Engineering Institute w l Academy of Engineering in 2009. In 2015 he became the head of Engineering Divisions
a
onathan Abir received a B.Sc. ( cum laude ) degree in mechanical engineering frongineering from the Technion, Haifa, Israel in 2013. He is currently working towar
niversity, Cranfield, UK. From 2008 to 2013, he was engaged in research and devis innovative and creative work.
aul Morantz is a Principal Research Fellow in the Precision Engineering Institut
athematical bases for the complex tool-path and metrology techniques utilized acrxperience in ultra-precision control, metrology and machine design.
tefano Longo is a Lecturer (assistant professor) in Vehicle Electrical and Electronic
e is the recipient of the 2011 IET Doctoral Dissertation prize for “significant achiemperial College London. He received his M. Sc . in control systems from the Univers
f Bristol, UK, in 2011. In November 2010, he was appointed to the position of resef 2012.
aul Shore was born in Chester, England, trained as machine tool designer prior
ngineering from Cranfield University, Cranfield, UK. From 1994 to 2002, he worke
recision production systems. In 2002 he returned to Cranfield as the McKeown Phere he span out Loxham Precision Limited. He was elected a Fellow of the Roya